摘要
研究PID控制器优化问题,现代工业控制过程中,由于许多被控对象受到干扰因素影响,具有高度非线性和不确定性,常规PID控制精度低,提出一种遗传算法、粒子群算法和RBF神经网络相融合的PID控制器设计方法(GA-PSO-RBF)。首先采用遗传算法选择PID控制参数初始值,然后采用粒子群算法优化RBF神经网络参数,采用优后的RBF神经网络辨识控制对象的输出对输入的变化灵敏度,最后采用单神经元对PID控制器进行在线性调整,得到理想的控制效果。仿真结果表明,GA-PSO-RBF神经网络PID控制器的超调量小,响应速度快,提高了系统的控制精度。
Study the optimization of PID controller. The paper put forward an PID controller design method, which integrates genetic algorithm, particle swarm algorithm and RBF neural network （GA -PSO -RBF） together. In this method, the genetic algorithm is used to select PID control parameter initial values, then the particle swarm algorithm to optimize RBF neural network parameters, and the optimized RBF neural network is used to identify the sensitivity of outputs and inputs of the control object. Finally, the single neuron is used for the on - ling linear tuning of PID controller to obtain ideal control effects. The simulation results show that, the GA - PSO - RBF PID neural network controller has the advanteges of small overshoot and fast response speed, and can improve the precision of control system.
出处
《计算机仿真》
CSCD
北大核心
2012年第12期180-183,共4页
Computer Simulation
关键词
参数优化
神经网络
遗传算法
粒子群优化算法
Cparameters optimization
Neural network
Genetic algorithm
Particle swarm optimization algorithm （ PSO ）
